Nitrate Reductase Assay for Rapid Detection of Isoniazid, Rifampin,Ethambutol, and Streptomycin Resistance in Mycobacteriumtuberculosis: a Systematic Review and Meta-Analysis

Ahmet Yilmaz Coban,a Aydin Deveci,b Ahmet Tevfik Sunter,c Anandi Martind

‹Ondokuz Mayis University, Medical School, Department of Medical Microbiology, Samsun, Turkeya; Ondokuz Mayis University, Medical School, Department of InfectiousDisease and Clinical Microbiology, Samsun, Turkeyb; Ondokuz Mayis University, Medical School, Department of Public Health, Samsun, Turkeyc; Laboratory ofMicrobiology, Faculty of Sciences, Department of Biochemistry and Microbiology, Ghent University, Ghent, Belgiumd

Colorimetric phenotypic tests recently gained interest because traditional primary drug susceptibility testing of Mycobacteriumtuberculosis isolates takes a long time. We used meta-analysis techniques to review the reliability and accuracy of the nitrate re-ductase assay (NRA), which is one of the most popular colorimetric methods to detect resistance to first-line drugs. Medline,PubMed, ISI Web, Web of Science, and Google Scholar were used to search for studies enrolled in the meta-analysis. The analysisincluded 35 studies for isoniazid (INH), 38 for rifampin (RIF), and 22 for ethambutol (EMB) and streptomycin (STR). Summaryreceiver operating characteristic (SROC) curves were applied to summarize diagnostic accuracy. The meta-analyses were per-formed by the use of Meta-DiSc software (version 1.4) and were focused on sensitivity and specificity values for measurements ofaccuracy. The pooled sensitivities were 96% for INH, 97% for RIF, 90% for EMB, and 82% for STR. The pooled specificities forINH, RIF, EMB, and STR were 99%, 100%, 98%, and 96%, respectively. The times required to obtain results were between 5 and28 days by the direct NRA and between 5 and 14 days by the indirect test. In conclusion, the present meta-analysis showed thatthe NRA is a reliable low-cost rapid colorimetric susceptibility test that can be used for the detection of multidrug-resistant(MDR) tuberculosis, including detection of EMB resistance. However, the test appears to have a relatively low sensitivity for STRand needs further improvement.

Tuberculosis (TB) is still an important cause of morbidity andmortality in developing countries (1). Recently, increasing in-

cidences of multidrug-resistant (MDR) and extensively drug-re-sistant (XDR) isolates have highlighted the importance of earlydiagnosis and treatment for control precautions (2). The treat-ment of patients infected with MDR and XDR isolates requiresprolonged use of secondary antituberculosis drugs. It is importantto diagnose and treat patients with MDR- and XDR-TB infectionsto prevent the spread of drug-resistant isolates (2, 3).

The proportion method (PM) is a conventional referencemethod that is used for the determination of primary antituber-culosis drug susceptibility. The CLSI recommends that thismethod be performed on Middlebrook 7H10 and 7H11 agars;however, it is commonly performed on Löwenstein-Jensen (LJ)medium (4, 5). The PM requires approximately 21 to 28 days forresults to be obtained. Therefore, the long period for detection ofresistant isolates leads to failure in the management of patientsinfected with MDR- and XDR-TB. Reference automated systems(Bactec 460 TB and Bactec MGIT 960 [Becton, Dickinson Diag-nostic Systems, Sparks, MD]) are available for overcoming thisproblem; however, the high costs and special requirements forthese devices impede their usage for early diagnosis of resistance,especially in countries with limited resources (6). This issue hasincreased the interest in colorimetric methods, which are rapid,reliable, inexpensive, and easily performed and interpreted.Among these phenotypic methods, the nitrate reductase assay(NRA) is one that has gained the most interest. The NRA can beused as a direct test on smear-positive sputum specimens or as anindirect test on Mycobacterium tuberculosis isolates grown fromconventional solid culture. The NRA is a technique based on thecapacity of M. tuberculosis to reduce nitrate to nitrite, which is

detected by adding the Griess reagent to the medium. In addition,this method is among the noncommercial drug susceptibilitymethods recommended by the WHO (3, 7).

We aimed to assess the accuracy of the NRA by a new system-atic review and meta-analysis, complementary to the previousones published (1, 3), for the determination of resistance to pri-mary antituberculosis drugs, including isoniazid (INH), rifampin(RIF), ethambutol (EMB), and streptomycin (STR), among M.tuberculosis isolated from both clinical specimens and sputumsamples.

MATERIALS AND METHODSLiterature search. Medline, PubMed, ISI Web, Web of Science, andGoogle Scholar were used to search for studies enrolled into the meta-analysis. The keywords used in our search were “nitrate reductase,”“Griess,” “nitrate test,” “colorimetric,” “Mycobacterium tuberculosis,”“tuberculosis,” “drug susceptibility,” “drug resistance,” “isoniazid,” “ri-fampin,” “ethambutol,” and “streptomycin.” Studies obtained fromsearching were enrolled in the analysis if they met inclusion criteria; nolanguage limitations were placed on the studies included.

Received 25 July 2013 Returned for modification 25 August 2013Accepted 8 October 2013

Published ahead of print 16 October 2013

Editor: G. A. Land

Address correspondence to Ahmet Yilmaz Coban, cobanay2003@gmail.com.

Supplemental material for this article may be found at http://dx.doi.org/110.1128/JCM.01990-13.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

doi:10.1128/JCM.01990-13

January 2014 Volume 52 Number 1 Journal of Clinical Microbiology p. 15–19 jcm.asm.org 15

on April 12, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

Study selection. We included studies that compared the NRA with areference standard method (including the proportion method on LJ me-dium or agar, Bactec 460-TB, or Bactec MGIT960), studies that detectedINH, RIF, EMB, and STR resistance, and studies that reported data onfalse-positive, true-positive, false-negative, and true-negative results. Theheterogeneity of the data was addressed by performing a subgroup anal-ysis with the NRA performed directly on sputum samples or indirectly onclinical isolates, in either solid or liquid medium.

Data extraction. Studies obtained by searches using defined keywordswere assessed by two independent reviewers. Disagreements between re-viewers were resolved by consensus. Titles and abstracts of studies wereread, and eligible ones were considered for data extraction. If study datawere found to be inconsistent by two independent reviewers, the studywas excluded from the meta-analysis. After that, data of the includedstudies were entered into the datasheet of Microsoft Excel software by onereviewer. Data extracted from the eligible studies included first author andyear of publication of the study, country in which the study was per-formed, reference method used in the study, number of samples (direct orindirect test), sensitivity, specificity, and time for obtaining results. Allextracted data were checked by a second independent reviewer. Sensitivitywas defined as the proportion of isolates correctly determined as resistantby use of the NRA compared with the reference method. Specificity wasdefined as the proportion of isolates correctly determined susceptible byuse of the NRA compared with the reference method.

Meta-analysis. Meta-analyses were performed by using Meta-DiScsoftware (version 1.4) (8) and were focused on the sensitivity and speci-ficity values for measurements of diagnostic accuracy. We created forestplots to estimate the accuracy of each test and receiver operating charac-teristic (ROC) curves that indicated the relationship between the true-positive results (TPR) and the false-positive results (FPR) of the test. TheQ* index values represent a summarization of the test performance, inwhich sensitivity and specificity are equal. The heterogeneity among stud-ies was analyzed by using the �2 and I2 tests for heterogeneity (interpretedas the percentage of the total variability due to true heterogeneity in a setof effect sizes) included in the Meta-DiSc program.

Quality assessment. The quality of individual studies was assessedusing criteria based on the Quality Assessment of Diagnostic AccuracyStudies (QUADAS) tools (9).

RESULTSData presentation. Cumulative data of all studies, includingnumbers of true-positive, false-positive, false-negative, and true-negative results, the AUC, and the Q* index values, are summa-rized in Table 1. Studies in which the NRA was used for the deter-mination of drug resistance, including INH, RIF, EMB, and STRresistance, are summarized regarding country in which the studywas performed, reference method used, number and type of sam-ples, sensitivity and specificity values, and time for obtaining re-sult in Tables S1 to S4 in the supplemental material. One study(10) had been performed in 5 different centers, and the results hadbeen evaluated individually in tables.

Detection of INH resistance. Thirty-five studies were re-trieved and included in the analysis. Positive sputum samples hadbeen evaluated by direct test in 8 studies, whereas bacterial isolateshad been evaluated by indirect test in the remaining 27 studies.Different reference methods had been used. The PM on LJ me-dium was used in 28 studies, the Bactec 460 TB in 3 studies, bothPM and Bactec MGIT 960, PM and Bactec 460 TB, or Bactec 460TB and Bactec MGIT 960 in 1 study, and PM on 7H10 agar in 1study.

Figures 1 and 2 illustrate the forest plots that we used to esti-mate the sensitivities and specificities of the NRA based on resultsof the 35 included studies. Sensitivities (Fig. 1; see also Table S1 inthe supplemental material) were 100% in 14 studies and �90% in19 studies. A sensitivity of 85% was found in one study and 87%in another one. Specificities (Fig. 2; see also Table S1) had been deter-mined as 100% in 21 studies and �93% in the remaining studies. TheSROC curve shows an AUC of 0.99 and a Q* index of 0.96, indicatinga high level of overall accuracy (Table 1; see also Fig. S1).

Detection of RIF resistance. Thirty-eight studies were in-cluded for RIF resistance. Positive sputum samples had beentested by direct test in 10 studies and by indirect test in 28 studies.The reference method used was PM on LJ medium in 30 studies,Bactec 460 TB in 3 studies, both PM and Bactec MGIT 960 or Bactec460 TB and Bactec MGIT 960 in 1 study, both PM and Bactec 460 TBin 2 studies, both Bactec MGIT 960 and Bactec 460 TB in 1 study, andPM on 7H10 agar in 1 study.

Figures 3 and 4 illustrate the forest plots and Fig. S2 in thesupplemental material illustrates the SROC of the 38 includedstudies. Sensitivities were 100% in 21 studies (Fig. 3; see also TableS2 in the supplemental material) and �90% in 14 studies andranged between 80 and 88% in 4 studies. Specificities had beendetermined as 100% in the majority of the studies (Fig. 4; see alsoTable S2). The SROC curve shows an AUC of 0.99 and a Q* index

TABLE 1 Cumulative data of all studies; numbers of true-positive, false-positive, false-negative, true-negative, AUC, and Q* values

Anti-TBdrug

No. ofstudies(D/I)a

No. of NRA studies (D/I) with the indicatedresultb

AUC Q*TP FP FN TN Total

INH 8/23 340/1027 11/12 10/46 667/1742 1028/2827 0.9933 0.9672RIF 10/24 298/948 6/5 10/30 1101/1865 1415/2848 0.9960 0.9759EMB 3/16 59/431 2/26 17/40 188/1597 266/2094 0.9820 0.9411STR 3/16 84/645 9/55 8/153 165/1293 266/2146 0.9788 0.9349

a D, direct test; I, indirect test.b TP, true positive; FP, false positive; FN, false negative; TN, true negative.

FIG 1 Forest plot of the sensitivities for INH assays. The point estimates ofsensitivities from each study are shown as circles. Error bars indicate 95%confidence intervals.

Coban et al.

16 jcm.asm.org Journal of Clinical Microbiology

on April 12, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

of 0.97, indicating a high level of overall accuracy (Table 1; see alsoFig. S2).

Detection of EMB resistance. Twenty-two studies were ana-lyzed for EMB resistance. Positive sputum samples had been eval-uated by direct test in only 3 studies. The reference tests were PM

in 18 studies, Bactec 460 TB in 2 studies, both PM and BactecMGIT 960 in 1 study, and PM on 7H10 agar in 1 study.

Figures 5 and 6 illustrate the forest plots and Fig. S3 in thesupplemental material shows the SROC curve of the same data.Sensitivities (Fig. 5; see also Table S3 in the supplemental material)were 100% in 8 studies and ranged between 88 and 98% in 8 otherstudies. For 6 studies, the sensitivities ranged between 55 and 75%.Specificities (Fig. 6) were 100% in 9 studies and �90% in 12 stud-

FIG 2 Forest plot of the specificities for INH assays. The point estimates ofspecificities from each study are shown as circles. Error bars indicate 95%confidence intervals.

FIG 3 Forest plot of the sensitivities for RIF analyses. The point estimates ofsensitivities from each study are shown as circles. Error bars indicate 95%confidence intervals.

FIG 4 Forest plot of the specificities for RIF analyses. The point estimates ofspecificities from each study are shown as circles. Error bars indicate 95%confidence intervals.

FIG 5 Forest plot of the sensitivities for EMB. The point estimates of sensitiv-ities from each study are shown as circles. Error bars indicate 95% confidenceintervals.

Nitrate Reductase Assay for M. tuberculosis

January 2014 Volume 52 Number 1 jcm.asm.org 17

on April 12, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

ies; only 1 study had a specificity of 75%. The SROC curve showsan AUC of 0.98 and a Q* index of 0.94, indicating a good level ofoverall accuracy (Table 1; see also Fig. S3).

Detection of STR resistance. Twenty-two studies were alsoincluded for STR resistance (see Table S4 in the supplementalmaterial), and positive sputum samples had also been evaluated bydirect test in only 3 studies. The reference tests were PM in 19studies, Bactec 460 TB in 2 studies, and both PM and Bactec MGIT960 in 1 study.

Figures 7 and 8 illustrate forest plots. There was considerableheterogeneity within studies, and sensitivities were very variablebetween studies. Eleven studies had sensitivities that ranged be-tween 92 and 100%, and 6 had sensitivities that ranged between 73

and 89% (Fig. 7). However, 6 studies had specificities that rangedbetween 12 and 67%. Specificities (Fig. 8) had been determined as100% in 8 studies and ranged between 78 and 99% for the otherstudies. The SROC curve shows an AUC of 0.97 and a Q* index of0.93, still indicating acceptable overall accuracy (Table 1; see alsoFig. S4 in the supplemental material).

Time for obtaining test results. The times required to obtaintest results for all drugs were between 5 and 28 days for direct NRAand between 5 and 14 days for indirect tests.

DISCUSSION

The aim of this study was to perform a meta-analysis of the NRA forthe detection of primary antituberculosis drug resistance. In the liter-ature there are two meta-analysis studies in which the NRA was ana-lyzed. The first one, performed in 2008, focused on the determinationof INH and RIF resistance and included 13 studies for INH and 14studies for RIF (3). The second meta-analysis evaluated the perfor-mance of four direct-testing methods, including the NRA, withanalyses of only 4 studies for the determination of RIF and INH sus-ceptibility (1). The present meta-analysis, in addition to the 2 meta-analyses reviewed previously, is the first one in which all primaryantituberculosis drugs were analyzed, including INH, RIF, EMB, andSTR, and it included 35 studies. Overall, the included studies achievedgood performance according to the analysis done with the QUADAStool. Statistical tests, such as funnel plots, are not recommended forthe detection of potential publication bias in meta-analyses of diag-nostic data (11). However, publication bias is possible in our meta-analysis because studies with poor performance are usually not pub-lished.

NRA had high sensitivity and specificity, especially for deter-mination of INH and RIF resistance, as shown in the previouslypublished meta-analyses (1, 3). Sensitivities and specificities hadbeen determined to be between 84.9 and 100% and 93 and 100%for INH and between 80 and 100% and 96.4 and 100% for RIF,respectively. Both sensitivity and specificity values had been deter-mined to be lower for EMB and STR. AUC and Q* index values forEMB and STR were calculated by use of SROC curve analyses andwere 0.98 and 0.90 for EMB and 0.97 and 0.90 for STR. For RIF,

FIG 6 Forest plot of the specificities for EMB. The point estimates of specific-ities from each study are shown as circles. Error bars indicate 95% confidenceintervals.

FIG 7 Forest plot of the sensitivities for STR. The point estimates of sensitiv-ities from each study are shown as circles. Error bars indicate 95% confidenceintervals.

FIG 8 Forest plot of the specificities for STR. The point estimates of specific-ities from each study are shown as circles. Error bars indicate 95% confidenceintervals.

Coban et al.

18 jcm.asm.org Journal of Clinical Microbiology

on April 12, 2020 by guest

http://jcm.asm

.org/D

ownloaded from

the NRA had been performed directly in bacillus-positive sputumsamples in 10 studies, and results were obtained in 5 to 28 days.The advantage of the direct test is that the results can be obtainedin a shorter time because it does not require time for bacterialisolation, which takes up to 2 to 4 weeks.

The NRA was usually performed in conventional LJ medium; ithad also been performed in liquid media (7H9-S) in 5 studies(12–16). The times required to obtain results were relativelyshorter in these studies compared with those in studies performedon LJ medium. Also, susceptibilities of INH and RIF were testedon blood agar media, and it has been suggested that blood agar canbe used for the NRA (17). NRA results are negative for M. tuber-culosis isolates which do not have nitrate reductase enzymes; how-ever, such isolates are rare.

The WHO published a report, “Noncommercial culture anddrug susceptibility testing methods for screening patients at riskfor multidrug-resistant tuberculosis.” This document indicatesthat the NRA can be used but it is not faster than other conven-tional phenotypic tests performed in liquid media (7). However,there is a gain of 2 to 4 weeks in direct susceptibility testing. If drugresistance is quickly detected, patients with MDR-TB will receivean effective drug therapy regimen and thus will be less contagiousto the community. The reduction in the time necessary to obtainsusceptibility results is of fundamental importance.

The present study demonstrates that NRA is reliable for sus-ceptibility testing of INH and RIF, the two most important drugsfor the treatment of tuberculosis, and the performance of this testcontinues to be good over time. It is well known that for the twodrugs EMB and STR, susceptibility is difficult to test even by con-ventional methods. Especially for STR susceptibility testing, NRAperformance was low. The cost of processing a sample using theGriess method is $4.80 (not including labor or capital costs) com-pared with $10.00 for conventional DST (18). Also, reagents areusually easy to procure and distribute. The NRA direct methodsignificantly shortened the time required to obtain test results, andwe believe that direct NRA has a strong potential for the directdetection of first-line drugs in resource-limited settings.

Adequate infrastructure for DST capacity under biosafety con-ditions is required. The advantage of the direct NRA is that with-out the need to handle cultured isolates, drug susceptibility can bedone with low biohazard risks.

In conclusion, the present meta-analysis showed that the NRA hasacceptable and reliable sensitivity and specificity for determination ofprimary antituberculosis drug resistance, with lower values for EMBand STR. It also revealed that the NRA significantly shortened thetime required for obtaining susceptibility test results. However, addi-tional studies are required to better establish the accuracy of NRAapplied to sputum, but preliminary studies included in this meta-analysis suggest that NRA may be useful for first-line resistance de-tection in sputum samples. Another important feature of the NRA isthat it yields the test results almost within the same period of time asexpensive rapid commercial tests.

ACKNOWLEDGMENT

We have no funding or conflicts of interest to declare.

REFERENCES1. Bwanga F, Hoffner S, Haile M, Joloba ML. 2009. Direct susceptibility

testing for multidrug resistant tuberculosis: a meta-analysis. BMC Infect.Dis. 9:67. http://dx.doi.org/10.1186/1471-2334-9-67.

2. Minion J, Leung E, Menzies D, Pai M. 2010. Microscopic-observation drugsusceptibility and thin layer agar assay for the detection of drug resistant tu-berculosis: a systematic review and meta-analysis. Lancet Infect. Dis. 10:688–698. http://dx.doi.org/10.1016/S1473-3099(10)70165-1.

3. Martin A, Panaiotov S, Portaels F, Hoffner S, Palomino JC, Angeby K.2008. The nitrate reductase assay for the rapid detection of isoniazid andrifampicin resistance in Mycobacterium tuberculosis: a systematic reviewand meta-analysis. J. Antimicrob. Chemother. 62:56 – 64. http://dx.doi.org/10.1093/jac/dkn139.

4. Clinical and Laboratory Standards Institute (CLSI). 2011. Susceptibilitytesting of Mycobacteria, Nocardia, and other aerobic actinomycetes. Ap-proved standard M24-A2, 2nd ed. Clinical and Laboratory Standards In-stitute, Wayne, PA.

5. Canetti G, Fox W, Khomenko A, Mahler HT, Menon NK, MitchisonDA, Rist N, Smelev NA. 1969. Advances in techniques of testing myco-bacterial drug sensitivity and the use of sensitivity tests in tuberculosiscontrol programs. Bull. World Health Org. 41:21– 43.

6. Martin A, Portaels F, Palomino JC. 2007. Colorimetric redox-indicatormethods for the rapid detection of multidrug resistance in Mycobacteriumtuberculosis: a systematic review and meta-analysis. J. Antimicrob. Che-mother. 59:175–183. http://dx.doi.org/10.1093/jac/dkl477.

7. World Health Organization. 2011. Noncommercial culture and drug-susceptibility testing methods for screening patients at risk for multidrug-resistant tuberculosis. World Health Organization, Geneva, Switzerland.http://whqlibdoc.who.int/publications/2011/9789241501620_eng.pdf.

8. Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A. 2006.Meta-DiSc: a software for meta-analysis of test accuracy data. BMC Med.Res. Methodol. 6:31. http://dx.doi.org/10.1186/1471-2288-6-31.

9. Whiting P, Rutjes AWS, Reitsma JB, Bossuyt PM, Kleijnen J. 2003. Thedevelopment of QUADAS: a tool for the quality assessment of studies ofdiagnostic accuracy included in systematic reviews. BMC Med. Res. Meth-odol. 3:25. http://dx.doi.org/10.1186/1471-2288-3-25.

10. Martin A, Montoro E, Lemus D, Simboli N, Morcillo N, Velasco M,Chauca J, Barrera L, Ritacco V, Portaels F, Palomino JC. 2005. Multi-center evaluation of the nitrate reductase assay for drug resistance detec-tion of Mycobacterium tuberculosis. J. Microbiol. Methods 63:145–150.http://dx.doi.org/10.1016/j.mimet.2005.03.004.

11. Flores LL, Steingart KR, Dendukuri N, Schiller I, Minion J, Pai M,Ramsay A, Henry M, Laal S. 2011. Systematic review and meta-analysisof antigen detection tests for the diagnosis of tuberculosis. Clin. VaccineImmunol. 18:1616 –1627. http://dx.doi.org/10.1128/CVI.05205-11.

12. Syre H, Phyu S, Sandven P, Bjorvatn B, Grewal HM. 2003. Rapidcolorimetric method for testing susceptibility of Mycobacterium tubercu-losis to isoniazid and rifampin in liquid cultures. J. Clin. Microbiol. 41:5173–5177. http://dx.doi.org/10.1128/JCM.41.11.5173-5177.2003.

13. Syre H, Valvatne H, Sandven P, Grewal HM. 2006. Evaluation of thenitrate-based colorimetric method for testing the susceptibility of Myco-bacterium tuberculosis to streptomycin and ethambutol in liquid cultures.J. Antimicrob. Chemother. 57:987–991. http://dx.doi.org/10.1093/jac/dkl054.

14. Affolabi D, Odoun M, Sanoussi N, Martin A, Palomino JC, Kestens L,Anagonou S, Portaels F. 2008. Rapid and inexpensive detection of mul-tidrug-resistant Mycobacterium tuberculosis with the nitrate reductase as-say using liquid medium and direct application to sputum samples. J. Clin.Microbiol. 46:3243–3245. http://dx.doi.org/10.1128/JCM.00083-08.

15. Coban AY, Uzun M, Akgunes A, Durupinar B. 2012. Comparativeevaluation of the microplate nitrate reductase assay and the resazurin mi-crotitre assay for the rapid detection of multidrug resistant Mycobacteriumtuberculosis clinical isolates. Mem. Inst. Oswaldo Cruz 107:578 –581. http://dx.doi.org/10.1590/S0074-02762012000500002.

16. Kumar M, Khan IA, Verma V, Kalyan N, Qazi GN. 2005. Rapid,inexpensive MIC determination of Mycobacterium tuberculosis isolates byusing microplate nitrate reductase assay. Diagn. Microbiol. Infect. Dis.53:121–124. http://dx.doi.org/10.1016/j.diagmicrobio.2005.05.011.

17. Coban AY, Tanriverdi Cayci Y, Deveci A, Akgunes A, Uzun M,Durupinar B. 2011. A rapid detection of multidrug-resistant Mycobac-terium tuberculosis by a nitrate reductase assay on blood agar. Mem.Inst. Oswaldo Cruz 106:378 –380. http://dx.doi.org/10.1590/S0074-02762011000300022.

18. Asencios L, Yale G, Yagui M, Quispe N, Taylor A, Blaya J, Contreras C,Cegielski P, Bayona J, Bonilla C, Shin S. 2008. Programmatic imple-mentation of rapid DST for Mycobacterium tuberculosis in Peru. Int. J.Tuberc. Lung Dis. 12:743–749.

Nitrate Reductase Assay for M. tuberculosis

January 2014 Volume 52 Number 1 jcm.asm.org 19

on April 12, 2020 by guest

http://jcm.asm

.org/D

ownloaded from